Window aggregations

Flink SQL uses window table-valued functions to support aggregations over time windows. These functions support several types of windows:

• tumbling
• hopping
• cumulating
• session

Session windows group events into sessions separated by gaps where the stream is idle for some minimum interval. The other window types all group events into windows of a fixed size relative to the time of day.

For windows relative to each event (rather than the time of day), see the module on OVER windows.

For a guided exploration into how both window and over aggregations work, see the accompanying exercise.

Documentation

• Windowing Table-Valued Functions (Windowing TVFs)
• Window Aggregation Queries